Near-field correction of CSAMT data based on Newton iteration method and GA method

Artificial source used in CSAMT studies has greatly enhanced the high frequency signal of the audio frequency bandwidth used in subsurface prospecting. On the other hand, it causes non-planar wave, known as near-field effect in CSAMT Cagniard apparent resistivity sounding curves. This is observed as a 45 inclination in low-frequency zone on bi-logarithmic plot of subsurface apparent resistivity. Hence, a method to correct for the CSAMT near-field effect based on Newton iteration method was first established. First, we derived an equation to calculate apparent resistivity from CSAMT data and solved it using Newton method to obtain corrected apparent resistivity. The accuracy of this method was validated using two-layer and three-layer theoretical models. In order to suppress both near-field effect and noise in data concurrently, we further introduced genetic algorithm (GA) inversion to the correction of CSAMT near-field by proposing an adaptive regularization factor to minimize the objective function, which is composed of whole-frequency error and model parameter gradient. The GA method was tested with a noisy data for effective removal of the near-field effect and suppression of noise in the data. When applied to real CSAMT data from Inner Mongolia, Newton iteration method and GA method were able to correct near-field effect effectively, thus displaying the accurate subsurface geo-electric structure.